Deep Learning-Based Multi-Label Tissue Segmentation and Density Assessment from Mammograms

Objectives: Breast cancer is the most commonly diagnosed type of cancer among women and a common cause of cancer-related deaths. Early diagnosis and treatment of breast cancer is critical in disease prognosis. Breast density is known to have a correlation with breast cancer. In recent years, there has been an increasing interest in the investigation of computer-aided methods for early diagnosis of breast cancer. In this study, a new fully-automated deep learning-based cascaded model was proposed for breast density assessment. In the first stage, the segmentation of adipose, fibroglandular, and pectoral muscle tissues from the digitized film mammograms of the Digital Database for Screening Mammography (DDSM) was investigated using various types of U-nets. Features extracted from the breast tissue segmentation predictions were then used to assess breast density in the second stage. Material and methods: 66 and 296 mediolateral oblique mammograms were selected from DDSM dataset for segmentation and breast density assessment systems, respectively. Different U-nets with varying number of layers and filters were implemented and the model having the highest performance was determined. U-net performance was investigated using categorical cross-entropy, Dice, Tversky, Focal Tversky, and logarithmic cosine-hyperbolic Dice loss functions. The performances of U-nets having different types of connections were investigated. The performances of U-nets having pre-trained weights from VGG16, VGG19, and ResNet50 networks in the encoding path were also investigated. Segmentation results were improved by using an image processing pipeline based on morphological operators. Segmentation performance was presented in terms of accuracy, balanced accuracy, intersection over union, and Dice's similarity coefficient (DSC) metrics. The segmentation system predictions were then used to estimate mammographic density using a machine learning pipeline by extracting features related to the fibroglandular tissue percentage. Results: Using ResNet50-U-net on the test data, average DSC scores of 82.71%, 73.39%, and 95.30% were obtained for adipose, fibroglandular, and pectoral muscle tissue segmentation, respectively. The mammogram segmentation results are 3%-12% better than the current state-of-the-art DSC in the literature when considering all of the foreground tissues concurrently. A breast density classification accuracy of 76.01% was achieved on a separate mammogram dataset, which is comparable to the recent studies in the literature. Conclusion: The proposed system can be used for automatic segmentation of mammogram into adipose, fibroglandular, and pectoral muscle tissues. The segmentation model enables the estimation of the fibroglandular-adipose tissue interface, which is recently found to be an important region for breast cancer investigations. The proposed fully-automatic breast density assessment system has a comparable performance to the ones in the literature.     

Probing the adaptive landscape using experimental islands: density-dependent natural selection on lizard body size

Anolis lizards in the Greater Antilles are thought to have diversified through natural selection on body size and shape, presumably due to interspecific competition and variation in locomotor performance. Here we measure natural selection on body size over three years and across seven replicate populations of the brown anole, A. sagrei. We experimentally manipulated an important component of the environment (population density) on several small islands to test the role of density in driving natural selection. Results indicate that the strength of natural selection was proportional to population density (r2 = 0.81), and favored larger body sizes at higher density, presumably owing to the enhanced competitive ability afforded by large size. Changes in the distribution of body size by selective releases of lizards to islands show that this effect did not arise by pure density dependence, since smaller individuals were disproportionately selected against at higher densities. We measured significant broad sense heritability for body size in the laboratory (h2 = 0.55) indicating that selection in the wild could have an evolutionary response. Our results suggest an important effect of population density on natural selection in Anolis lizards.     

c-Src is a PDZ interaction partner and substrate of the E3 ubiquitin ligase Ligand-of-Numb protein X1

The very C-terminus of c-Src is a ligand for PDZ domains. In a screen for PDZ domains that interact with c-Src, we identified one of the PDZ domains of the Ligand-of-Numb protein X1 (LNX1), a multiple PDZ domain scaffold and RING type E3 ubiquitin ligase. We demonstrate that the interaction of c-Src with LNX1 depends on the C-terminal PDZ ligand of c-Src. Furthermore, we show that c-Src phosphorylates LNX1. Moreover, c-Src itself is ubiquitinated by LNX1, suggesting an interdependent regulation of c-Src and LNX1.     

The effects of alternating tangential flow (ATF) residence time,hydrodynamic stress,and filtration flux on high-density perfusion cell culture

In this study, we investigated the effects of alternating tangential flow (ATF) cell separation on high-density perfusion cultures. We have developed methods to estimate theoretical residence times of cells in the ATF system and discovered that long residence times (above 75 s) correlate with decreased growth, metabolism, and productivity. We have calculated energy dissipation rates in the ATF transfer line and filter and empirically studied the impacts of increased exchange rates on cell culture, determining that increased hydrodynamic stress can lead to decreased cell size, lactate production, and specific productivity. Finally, we have conducted experiments to understand the relationship between filtration fluxes and ATF membrane fouling, finding that at fluxes above 60 L·m^–2·day ^–1, protein sieving coefficients see significant rates of decrease (greater than 1% per day). While most of these studies have been conducted with one cell line at one target viable cell density (40 million cells/ml), the general, directional knowledge arising from this study should be applicable to other conditions and programs, ultimately leading to more robust and well-designed perfusion processes.     

Assessment of the chitosan-functionalized graphene oxide as a carrier for loading thioguanine,an antitumor drug and effect of urea on adsorption process: Combination of DFT computational and molecular dynamics simulation studies

In this study, the interaction thioguanine (TG) anticancer drug with the functionalized graphene oxide (GO) nanosheet surface is theoretically studied in both gas phase and separately in physiological media using the density functional theory (DFT) calculations. DFT calculations indicated the adsorption and solvation energies are negative for f-GONS/TG complexes which propose the adsorption process of TG molecule onto the f-GONS surface is possible from the energetic viewpoint. QTAIM calculations confirm the nature of partially covalent-partially electrostatic between drug and nanosheet. These results are sorely relevant that an approach for loading of TG molecule is the chemical modification of GO using covalent functionalization which can serve as a nanocarrier to load drug molecules. Moreover, to understand the effect of urea on the nature of the interaction between TG and f-GONS, molecular dynamics (MD) simulation was employed. The results indicated that in the presence of urea the adsorption process gets affected and leads to instability of system, while the affinity of the TG for adsorption onto GO surface is increased in pure water.

Communicated by Ramaswamy H. Sarma     

Impact of apolipoprotein A1- or lecithin:cholesterol acyltransferase-deficiency on white adipose tissue metabolic activity and glucose homeostasis in mice

High density lipoprotein (HDL) has attracted the attention of biomedical community due to its well-documented role in atheroprotection. HDL has also been recently implicated in the regulation of islets of Langerhans secretory function and in the etiology of peripheral insulin sensitivity. Indeed, data from numerous studies strongly indicate that the functions of pancreatic β-cells, skeletal muscles and adipose tissue could benefit from improved HDL functionality. To better understand how changes in HDL structure may affect diet-induced obesity and type 2 diabetes we aimed at investigating the impact of Apoa1 or Lcat deficiency, two key proteins of peripheral HDL metabolic pathway, on these pathological conditions in mouse models. We report that universal deletion of apoa1 or lcat expression in mice fed western-type diet results in increased sensitivity to body-weight gain compared to control C57BL/6 group. These changes in mouse genome correlate with discrete effects on white adipose tissue (WAT) metabolic activation and plasma glucose homeostasis. Apoa1-deficiency results in reduced WAT mitochondrial non-shivering thermogenesis. Lcat-deficiency causes a concerted reduction in both WAT oxidative phosphorylation and non-shivering thermogenesis, rendering lcat^?/? mice the most sensitive to weight gain out of the three strains tested, followed by apoa1^?/? mice. Nevertheless, only apoa1^?/? mice show disturbed plasma glucose homeostasis due to dysfunctional glucose-stimulated insulin secretion in pancreatic β-islets and insulin resistant skeletal muscles. Our analyses show that both apoa1^?/? and lcat^?/? mice fed high-fat diet have no measurable Apoa1 levels in their plasma, suggesting no direct involvement of Apoa1 in the observed phenotypic differences among groups.     

Influence of crucifer cropping system on the parasitism ofPlutella xylostella (Lep., Yponomeutidae) byCotesia plutellae (Hym., Braconidae) andDiadegma semiclausum (Hym., Ichneumonidae)

Field experiments were conducted to study the influence of cabbage monoculture and mixed cropping on the parasitism of diamondback moth,Plutella xylostella (L.), a destructive pest of all crucifers, by 2 larval parasites,Diadegma semiclausum Hellén andCotesia plutellae Kurdjumov. There was no significant difference in parasitism by either species whether cabbage was planted in insecticide-free monoculture or in mixed cropping with 8 noncrucifers which were sprayed twice a week with chemical insecticides mevinphos, methamidophos and permethrin. Population ofP. xylostella increased as the cabbage plants grew older. Parasitism byC. plutellae was higher soon after cabbage transplanting but decreased as the plants grew older. Parasitism byD. semiclausum was very low soon after cabbage planting but increased as the plants grew older. A significant negative correlation was found betwen parasitism byC. plutellae andD. semiclausum. In a caged field study where only one parasite species was used in an individual cage, parasitism ofP. xylostella by both species decreased as theP. xylostella population increased. This is believed to be due to the absence of competition between the two parasites inside the cage. There was no relationship between host-plant age and parasitism ofP. xylostella larvae by either parasite species.     

Flax (Linum usitatissimum L.) depends on arbuscular mycorrhizal fungi for growth and P uptake at intermediate but not high soil P levels in the field

The contribution of indigenous arbuscular mycorrhizal fungi (AMF) to growth and phosphorus (P) uptake by oilseed flax (Linum usitatissimum L.) was examined in two field experiments covering soil P levels from 20–86 mg kg-1 NaHCO3-extractable P. The fumigant dazomet was applied to the soil in half of the plots to obtain control plants with reduced mycorrhiza formation. An extensive AMF colonization of up to 48% of the root length was established in untreated soil of both experiments, although P fertilization reduced colonization to 28–39% at the latest harvests. Fumigation markedly decreased or totally prevented AMF colonization throughout the experiments. Root growth responded to fumigation by increased total and specific root length. Shoot P uptake was decreased by fumigation at soil P levels lower than ca. 50 mg kg-1 whereas shoot growth was reduced by fumigation at soil P levels lower than ca. 40 mg kg-1. The effects of fumigation were ascribed to the suppression of mycorrhiza formation. The effect of the AMF increased with decreasing soil P levels. Phosphorus inflow through roots (based on shoot P uptake) was reduced more strongly by fumigation than total P uptake. The P inflow through fungal tissue in roots was estimated to 4 × 10-14 mol P cm-1 s-1. We conclude that AMF are essential to flax growth at soil P levels below ca. 40 mg P kg-1, which is representative of the conditions under which most flax is grown.     

我国东、南沿海地区AM真菌群落生态分布研究

在我国东、南沿海地区AM真菌资源调查的基础上研究了AM真菌的群落生态分布和特征。证明球囊霉属在该地区的出现频度和相对多度都最高,其次是硬囊霉属。该区内南四省AM真菌丰度、密度和物种多样性指数均比北三省高。聚类分析结果和上述相似,说明我国东、南沿海各省AM真菌的丰度、频度、密度、多度和多样性特征由北向南呈规律性变化。     

Spatial ecology of wolverines at the southern periphery of distribution

Wolverines (Gulo gulo) in the conterminous United States have experienced range contraction, are uncommon, and have been designated as warranted for protection under the United States Endangered Species Act. Data from the southern edge of the wolverine's circumpolar distribution is sparse, and development of effective conservation strategies would benefit from a more complete understanding of the species' ecology. We captured and radio-monitored 30 wolverines in the Greater Yellowstone Ecosystem (GYE), tested for seasonal habitat selection by elevation band, and examined a suite of spatial characteristics to clarify our understanding of the wolverine's niche. Wolverines in GYE selected for areas >2,600 m latitude-adjusted elevation (LAE; n = 2,257 wolverine locations [12 F, 6 M]). Wolverines avoided areas <2,150 m LAE, including during winter when the vast majority of ungulates are pushed to these elevations by deep snow. Wolverine home ranges were large relative to body size, averaging 303 km² for adult females and 797 km² for adult males (n = 13 [8 F, 5 M] and 33 wolverine-years). Resident adults fit with Global Positioning System (GPS) collars used an area >75% the size of their multi-year home range in an average of 32 days (n = 7 [5 F, 2 M]). Average movement rates of 1.3 km/2-hr indicated that both sexes move distances equivalent to the diameter of their home range every 2 days or the circumference of their home range in <1 week (n = 1,329 2-hr movements, n = 12 individuals [7 F, 5 M]). This capability for movement, the short time-frame over which home ranges were developed, and a lack of home range overlap by same sex adults ( , 90% CI = 0.0–4.8%, n = 22 pairs) suggested territoriality. We estimated wolverine density to be 3.5/1,000 km² of area >2,150 m LAE (95% CI = 2.8–9.6). Dispersal movements extended to at least 170 km for both sexes (n = 5 F, 2 M). At the southern edge of distribution, where suitable and unsuitable conditions exist in close proximity, wolverines selected high-elevation areas near alpine tree-line where a mix of forest, meadow, and boulder fields were present, deep snow-cover existed during winter, and low temperatures near freezing can occur throughout the year. Persistence in these areas where the growing season is brief requires large home ranges that are regularly patrolled, a social system that provides exclusive access to resources, and low densities. These characteristics, along with low reproductive rates, are prevalent throughout the species range, indicating that wolverines are specialists at exploiting a cold, unproductive niche where interspecific competition is limited. The vulnerability inherent in occupying this unproductive niche was likely influential in previous declines within the conterminous United States and will remain a factor as wolverines encounter modern human influences. Conserving wolverines in the conterminous United States will require collaborative management over a large geographic scale. © 2011 The Wildlife Society.